Call lm with a model matrix instead of a formula

Question

I want to fit a linear model in R using lm to obtain coefficient estimates and p-values + p-value for total model fit (ANOVA-like), so basically the output from summary.lm.

The problem is that I want to use my own model matrix, instead of specifying it using a formula when calling lm.

I've tried using the underlying lm.fit function which allows for this, but then I loose the convenience of the summary.lm function, and I don't want to recalculate all the test statistics again myself.

Is there a way to 'trick' lm and give it a model matrix rather than a formula?

Thanks!

Answer 1

Here is an example using the built in BOD data frame:

# inputs
demand <- BOD$demand
mm <- model.matrix(~ Time, BOD) # model matrix

summary(lm(demand ~ mm + 0))

or

summary(lm(demand ~. + 0, as.data.frame(mm)))